CentOS - Oct 2018 - What are the differences between systemd and non-systemd Linux distros?

On Tue, Oct 16, 2018 at 09:25:15AM -0500, Valeri Galtsev
wrote:
> Hoping to not offend proponents of systemd/firewalld...
Perhaps if you weren't spreading misinformation, we wouldn't be
offended? 
> Linux kernel is already containing chunks of code related to
> systemd/firewalld and friends. One can disable stuff during kernel build,
> but the result it still is not like the result of building kernel before
the
> existence of systemd/firewalld.
None of this is true.  It's true that systemd uses some Linux-only
features like cgroups, but I was using those features well before
systemd came around.  And firewalld uses Linux only specific features
too -- it manages the NETFILTER rules which is a linux-specific
project.  The only thing that seems to be in common is that they are
both projects that end with 'd'.  I suppose you're going to start
claiming that SSHd, HTTPd and NTPd are up to no good.
> Also, it is likely that at some point
> systemd-free Linux distribution(s) may fade away.
There was already a move away from SysV init before systemd was
introduced, heck RHEL6/CentOS6 used Upstart instead of SysV.  There
are always going to be projects with a diverse set of tools, it just
depends on how many people care about it.  Turns out, not that many
people care about maintaining a SysV init (or other init) distro.  

-- 
Jonathan Billings <billings at negate.org>

On 10/16/2018 10:27 AM, Jonathan Billings wrote:
> On Tue, Oct 16, 2018 at 09:25:15AM -0500, Valeri Galtsev wrote:
>
>> Also, it is likely that at some point
>> systemd-free Linux distribution(s) may fade away.
> There was already a move away from SysV init before systemd was
> introduced, heck RHEL6/CentOS6 used Upstart instead of SysV.  There
> are always going to be projects with a diverse set of tools, it just
> depends on how many people care about it.  Turns out, not that many
> people care about maintaining a SysV init (or other init) distro.
>
I'm not sure that that necessarily follows. Among RH-ecosystem 
distributions, and specifically RHEL derivatives, there's a barrier to 
the usefulness of smaller projects given that a large chunk of the users 
need binary-compatible commercial equivalents, or at least vaguely 
commercially supported ecosystems. We're long past the days where WBEL 
and other hobbyist projects can probably gain traction. Those RHEL 
alternatives that do exist either have a long history (CentOS, even 
before the RH deal), or are supported by large entities: the government 
(SL, before it became more or less congruent with CentOS), a multi 
billion dollar company (OEL), or a trillion dollar company (AWS). SuSE 
Enterprise might be the best counter example here.

Also, while EL6 did move from original init to upstart, that's somewhat 
beside the point. Almost none of the advanced features from upstart were 
used, and - crucially - the startup sequence was still handled with 
grokkable, imperative scripts. The jump from EL6->EL7 was night and day 
compared to EL5->EL6.

-jc

On 16/10/2018 19:21, Japheth Cleaver wrote:
> I'm not sure that that necessarily follows. Among RH-ecosystem
> distributions, and specifically RHEL derivatives, there's a barrier to
> the usefulness of smaller projects given that a large chunk of the
> users need binary-compatible commercial equivalents, or at least
> vaguely commercially supported ecosystems. We're long past the days
> where WBEL and other hobbyist projects can probably gain traction.
> Those RHEL alternatives that do exist either have a long history
> (CentOS, even before the RH deal), or are supported by large entities:
> the government (SL, before it became more or less congruent with
> CentOS), a multi billion dollar company (OEL), or a trillion dollar
> company (AWS). SuSE Enterprise might be the best counter example here.
>
> Also, while EL6 did move from original init to upstart, that's
> somewhat beside the point. Almost none of the advanced features from
> upstart were used, and - crucially - the startup sequence was still
> handled with grokkable, imperative scripts. The jump from EL6->EL7 was
> night and day compared to EL5->EL6.
Not that I disagree with the thrust of what you are saying but it seems
to me that SUSE is not so much a counter example. The SUSE subsidiary of
Micro Focus is, in and of itself, a multi-billion dollar company. It was
valued at $2.535 billion when its sale to EQT Partners was agreed
earlier this year.

It seems to me that what you say in your first paragraph above applies
not just to RH-ecosystem Linuxes but probably to all corporate-focussed
ones in both the RH and SUSE ecosystems.

It's mainly the Debian world where it seems to me that there is still
room for smaller entrants (including at least one healthy non-systemd one).



-- 
Mark Rousell